Yoshua Bengio, a leading figure in artificial intelligence (AI) and one of its "godfathers, " alongside fellow researchers Yann LeCun and Geoffrey Hinton, received the prestigious Turing Award in 2018 for their innovations in artificial neural networks and deep learning. However, Bengio is increasingly concerned about the rapid development and adoption of AI technologies, which he believes could pose significant risks to society. As the chair of the International Scientific Report on the Safety of Advanced AI, representing 30 nations and the EU, he advocates for caution and the implementation of regulations. In a recent interview at the HowTheLightGetsIn Festival, Bengio expressed worry over catastrophic consequences if AI development continues unchecked, particularly for future generations. He feels a personal responsibility to mitigate these potential dangers and emphasizes the need for technical and political solutions to ensure AI systems do not harm humanity. Bengio identified critical short- and long-term risks of AI, including the misuse of persuasive AI in political campaigns and the potential for superintelligent machines to escape human control or be weaponized by malicious actors.

He noted that while governmental responses, like the EU's AI act and Biden's executive order, are encouraging, they still lack the enforcement needed to protect the public adequately. Despite some technological advancements, such as improved unsupervised learning capabilities, there remain challenges, including high data requirements and risks associated with AI becoming overly dependent on synthetic data generation. Moreover, he highlighted concerns about the environmental impact of large-scale AI training, which contributes to increased fossil fuel consumption. Bengio diverged from the debate over artificial general intelligence (AGI) by focusing on the dangerous capabilities that AI could possess, regardless of whether those capabilities constitute AGI. He urged for more comprehensive action from both scientists and politicians to mitigate risks, acknowledging that current regulations are insufficient to protect against catastrophic outcomes. In conclusion, Bengio believes that while innovation in AI must continue, it needs to be done responsibly to prevent market forces from driving potentially destructive technologies without oversight. He advocates for international cooperation and treaties to manage the development of AI in a way that ensures shared safety and benefits for everyone.